Eighty days on Brightwater Health, €150,000, and the five weeks after. With a CRM connector, ChatGPT or Claude can read the deal and answer questions about it, and its answers here are good. It answers when someone asks. Step through the deal and see who asked, and when.
Forecast pipeline, qualify leads, and update deal stages in secondssource
Where the assistant is strong: asked the right question, it gives a good answer, and the post-mortem it writes after the close is excellent. Nudge runs inside Claude too, through its own connector.
We have not run ChatGPT or Claude on your CRM on this deal. The left side is our model of the approach it describes, in the words quoted here; its real output will differ. Brightwater Health is a composite. The account is invented, the sequence is not. No customer data is on this page.
11 events, in order. At each one, both sides say what they make of the deal and what they would do about it. Step forward and watch where the two readings part.
Not asked.
No answer given.
Do this: Nothing, until someone asks.
They agreed to a workshop with the ops team on the 12th.
A dated step, and both sides own it. A normal start.
Do this: Ask Priya to bring whoever signs, or to name them.
Not asked.
No answer given.
Do this: Nothing, until someone asks.
Priya will send three months of usage data by the 20th.
She owns it and she put the date on it herself. The best kind of promise.
Do this: Put the 20th in the recap, and ask who reviews the business case once the data is in.
Not asked.
No answer given.
Do this: Nothing, until someone asks.
Kept, a day early.
Her first promise landed. Her word now carries weight on this deal.
Do this: Build the business case on her numbers, and ask for the review date with the CFO.
Not asked.
No answer given.
Do this: Nothing, until someone asks.
Ours. We sent a document.
Our own effort. It moves nothing by itself.
Do this: Offer to walk Mark, the CFO, through it before he reads it alone.
Asked: “how is Brightwater going?” It reads the CRM: Proposal stage, activity six days ago, next step “procurement will be in touch”.
It looks on track.
Do this: Wait for procurement, and follow up next week.
Procurement will contact us. No date, no name.
The first promise on this deal with no owner and no day. Not a failure yet.
Do this: Ask Priya who in procurement, and by when.
Not asked. The CFO's decline is in the calendar, but nobody put the question.
No answer given.
Do this: Nothing, until someone asks.
None. Mark declined and offered no other slot.
The only person who signs closed the door, and nobody on their side owes us anything dated. This is the week they checked out.
Do this: Ask Priya plainly whether the budget is still there, and offer Mark fifteen minutes on his own terms.
Not asked.
No answer given.
Do this: Nothing, until someone asks.
They want our answers in their grid. Nothing offered back.
A grid on day 35, with the signer absent, is a reference quote being collected.
Do this: Answer it in exchange for a meeting with Mark, or treat the deal as lost.
Asked again. It finds the “on hold” email and the CFO's decline three weeks earlier.
On hold. The CFO has not engaged since the decline.
Do this: Re-engage the CFO before the quarter ends.
None.
On hold, with no date to come off hold.
Do this: Take it out of the quarter. Ask what would take it off hold, and when.
Not asked.
No answer given.
Do this: Nothing, until someone asks.
None.
Evaluating what, against whom, by when: none of it is on the record.
Do this: Say it in the forecast: not this quarter.
Asked: “why did we lose Brightwater?” It reads the whole record: the CFO declined 56 days before close and never came back, procurement opened a comparison grid a week later, and the deputy put it on hold.
Budget is the logged reason. The turn was the CFO's decline.
Do this: Look at CFO engagement earlier in similar deals.
Nothing left to test.
Budget was the word. The record says the CFO closed the door on day 42, and procurement started comparing on day 35.
Do this: Log day 42 as the turn, and look for the same week in the deals still open.
Not asked.
No answer given.
Do this: Nothing, until someone asks.
Nothing left to test.
Her answer fits the record: the CFO never saw a case strong enough to make room for it.
Do this: Use her answer to sharpen the CFO case in the deals still open.
The difference is who starts the conversation. The assistant waits for a question, and answers from what the CRM holds that day. Nudge watches every deal every week against how the company's buyers have bought before, and brought up day 42 without being asked.
One test, run on the only thing that predicts anything: what the buyer agreed to do next.
The next thing that has to happen is theirs, not ours.
There is a day on which it is late. Without one, nobody notices it did not happen.
This person’s last promises actually arrived.
It gets us to someone we could not reach before.
The person promising it has the standing to do it.
Each page puts one approach next to Nudge on the same deal or the same pipeline. Where the other approach is right, the page says so.
The original comparison: a CRM with a model on top, on one deal.
In thirty minutes on your own deals, before the board, we show you the week each slipped deal's buyer checked out, and which open deals will really close.